Arca Computer

Issue Scout is an AI agent that finds the best first issues in any GitHub repo and writes the fix plan for you. Live at https://issuescout.arca.computer Contributing to open source starts with a hard question: which of hundreds of open issues can a newcomer actually fix? Enter any owner/name GitHub repo and an agent built on Meta's Muse Spark fetches the open issues, scores each one 0–10 for newcomer-friendliness with a short reason, and picks the top three. Then it goes beyond a single prompt. For each pick, the model decides which tools to call: it browses the repository file tree, reads the README, and opens source files through the GitHub API. Only then does it write a concrete plan with the files likely involved, a step-by-step approach, and the tests to add. A streaming progress log shows every agent step and tool call in a clean single-page UI. It runs as a Cloudflare Worker at issuescout.arca.computer, and it also runs locally with one command (./run.sh). It helps new contributors get unstuck and helps maintainers grow their contributor base. Everything was built during the hackathon on Nov 4, 2026.